Contents lists available at ScienceDirect

Journal of Environmental Chemical Engineering

journal homepage: www.elsevier.com/locate/jece

Synthesis of CeO2/TiO2 nanotubes and heterogeneous photocatalyticdegradation of methylene blue

Le Thi Thanh Tuyena,c, Dao Anh Quangb, Tran Thanh Tam Toana, Truong Quy Tungb,Tran Thai Hoaa, Tran Xuan Maua, Dinh Quang Khieua,⁎

aUniversity of Sciences, Hue University, Hue City, 530000, Viet NambHue Industrial College, Hue City, 530000, Viet Namc Le Quy Don Gifted High School, Danang City, 550000, Viet Nam

A R T I C L E I N F O

Keywords:CeO2/TiO2 nanotubesPhotocatalytic activityVisible lightBox–Behnken design

A B S T R A C T

In this study, the preparation of CeO2/ TiO2 nanotubes (CeO2/TiO2–NTs) is demonstrated using the hydro-thermal method. The conditions for the synthesis were optimised using the Box–Behnken design. The samplesobtained were characterised by means of X-ray diffraction, high-resolution transmission electron microscopy,energy dispersive X-ray spectroscopy, transmission electron microscopy, scanning electron microscopy, X-rayphotoelectron spectroscopy, ultraviolet-visible diffuse reflectance spectroscopy, nitrogen adsorption/desorptionisotherms, and Raman spectroscopy. The photocatalytic behaviour under visible light and kinetics of the CeO2/TiO2–NTs catalyst via methylene blue degradation were addressed. The results showed that the introduction ofCeO2 into TiO2–NTs enhanced the photocatalytic activity in the visible light region. CeO2/TiO2–NTs were stableand potential as a visible light photocatalyst for the organic substances degradation in aqueous solutions.

1. Introduction

TiO2 is classified as a semiconductor, widely used in photochemicaltechniques to decompose numerous kinds of toxic organic contaminantsbecause of its outstanding features. TiO2 is a low-cost non-toxic com-pound with high chemical durability, high photochemical stability andbiological inertness [1,2]. Several oxide nanostructures of TiO2 havebeen explored, including nanoparticles (0D) [3], nanowires (1D) [4],nanorods (1D) [5], nanofibers [6], and nanosheets (2D) [7]. The uni-form dispersion of nanoparticles can be achieved in a liquid medium viaelectrostatic and steric stabilization. However, when the nanoparticlesare consolidated into solid materials, the aggregation between the na-noparticles becomes very strong because the van der Waals attraction isinversely proportional to the particle size [8]. When the aggregates arelarge and dense, only the primary particles near the surface region ofthe secondary particles contribute to the catalytic reaction and theinner part becomes inactive. Under this configuration, a high catalyticactivity cannot be achieved. The 1D nanostructures such as nanowires,nanorods, and nanotubes with a less agglomerated configuration havebeen used to improve the catalytic activity [4,7]. When photocatalytictitania (TiO2) absorbs ultraviolet radiation from a light source, anelectron of the valence band of titanium dioxide becomes excited. The

excess energy of the excited electron promotes the electron to theconduction band of titanium dioxide, therefore creating a negative-electron (e–) and positive-hole (h+) pair. The positive hole will oxidizewater molecules to form hydrogen gas and hydroxyl radicals. The ne-gative electron will reduce oxygen molecules to form a superoxideanion. This cycle continues when the light is available. As a result, thesefree radicals will decompose the organic contaminants [9,10]. Incomparison with nanoparticles, TiO2 nanotubes (denoted as TiO2-NTs)possess photocatalytic features. Depending on the method of synthesisutilised, their preeminent features are: massive surface (up to 478m2/g), great volume of capillary (up to 1.25 cm3/g) [11,12], capacity oftransferring electrons from long distances [13], capacity of ion ex-change [14], and noticeable capacity of absorbing light as a result of theconsiderable ratio between the length and the diameter of the tubes[15]. The transfer of the charge carriers along the length of the titaniananotubes has been considered to prevent the fast recombination ofphotoexcited electrons and positive holes [16,17]. Such unique char-acteristics of TiO2-NTs result in enhanced photocatalytic activities andrender them excellent candidates for photocatalytic applications. As forpreparation techniques, the hydrothermal process is considered as themost effective to synthesise TiO2-NTs with their large-sized capillaryand nanotube unique structure [11,18–21]. This is the simplest method,

https://doi.org/10.1016/j.jece.2018.09.022Received 13 June 2018; Received in revised form 13 August 2018; Accepted 16 September 2018

⁎ Corresponding author.E-mail address: dqkhieu@hueuni.edu.vn (D.Q. Khieu).

Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

Available online 19 September 20182213-3437/ © 2018 Elsevier Ltd. All rights reserved.

T

as it is easy to carry out without the requirement of moulds or bases.Furthermore, the output nanotubes have their homogeneity with anoticeable surface and porosity.

To improve the photocatalytic efficiency of TiO2, besides morpho-logical modification attempts, wide employment of binary oxides basedon TiO2 has yielded a number of marked results. Binary oxide systemssuch as TiO2–Fe2O3 [22], TiO2–SnO2 [23], TiO2–Cu2O [24], TiO2–CeO2

[25], and TiO2–WO3 [26] have been found to exhibit a much higherphotocatalytic activity under visible light illumination than pure TiO2

by accelerating the charge separation on the TiO2 surface and extendingthe energy range of photoexcitation. Being one of the rare-earth metaloxides, CeO2 has attracted a considerable attention due to its specialelectron orbital structure, unique optical and Ce3+/Ce4+ redox beha-viour, high thermal stability and large oxygen-storage capability. Sur-face defects such as oxygen vacancies working as electron traps canimpede e–/h+ recombination, and the 4f electron configuration canenhance the electron transfer from the adsorbed dye to oxygen species.Therefore, introducing CeO2 to TiO2 nanostructures could favour theimproved separation of the electron-hole pairs, thus enhancing thephotocatalytic performance of CeO2–TiO2 catalysts [27–30].

In practice, the influence of different synthesis parameters on theperformance of the obtained photocatalysts is crucial. However, most ofthe published results concerning the relationship between the photo-catalytic capacity of CeO2–TiO2 catalysts and various related para-meters followed the conventional method [2,31,32,33,34]. Such amethod is carried out by monitoring the effect of one factor at a time onan experimental response while maintaining other factors at a fixedlevel. This technique, therefore, does not consider the interactive effectsof the variables and requires considerable resources including time andexpenses for a large number of variables [35], whereas the optimisationof the procedures of synthesis by using multivariate statistic techniquesallows the simultaneous variation of many experimental factors. Themultivariable optimisation has been considered to be faster, moreeconomical and effective in identifying the best conditions and max-imising desirable responses in any experimental subjects. However, tothe best of our knowledge, there are only a few reports available in theliterature regarding the optimization of the CeO2–TiO2 preparationconditions in the photocatalytic degradation [1,36].

In the present work, the effective variables of the preparation ofCeO2/TiO2-NTs) for the photocatalytic degradation of methylene blue(MB) under visible light were optimised utilising the Box–Behnkendesign (BBD) of response surface methodology (RSM) for the maximalphotocatalytic yield. The photocatalytic behaviour and kinetics studiesof the photocatalytic degradation of methylene blue (MB) were in-vestigated.

2. Experimental

2.1. Materials

Cerium (III) nitrate hexahydrate (Ce(NO3)3.6H2O, Merck, Germany,99%), titanium (IV) oxide, anatase (TiO2, DaeJung, Korea, 98%), me-thylene blue (MB) (C16H18ClN3S, Merck, Germany), P25 (TiO2,Degussa, Germany, 99,5%); NaOH, NH3.H2O (Sinofarm ChemicalReagent Co., Ltd., China, 25%–28%) and HCl (Sinofarm ChemicalReagent Co., Ltd., China) were used as received without further pur-ification. Methylene blue (C16H18ClN3S, Merck, 99%) was used as a dyemodel. The molecular formula is presented in Fig. 1.

2.2. Preparation of CeO2/TiO2–NTs

TiO2–NTs were synthesised by using a hydrothermal route, whichhas been reported in detail in some literature [17,37–39] (see Fig. 2). Inbrief, 3 g of TiO2 powder was dispersed in 70mL of a 10M NaOH so-lution and the mixture was then sonicated continuously for 30min.Afterwards, the suspension was transferred to a 100mL Teflon-linedautoclave and hydrothermally treated at temperatures 140, 160, and180 °C for a specific time (18, 20 and 22 h). The suspension was thenfiltered, and the powder was washed several times with distilled waterand 0.1M HCl until the pH of the leaching water reached 7, followed byair-drying at 100 °C. Thus, TiO2–NTs were obtained. The CeO2/TiO2–NTs catalyst was prepared with the impregnation method de-scribed by Zhao et al. [40] as follows: 0.8 g of the as-prepared TiO2–NTswas dispersed in a specific amount of the aqueous solution of 0.02M Ce(NO3)3 under magnetic stirring for 2 h at room temperature. Simulta-neously, a sufficient amount of NH3.H2O solution was added drop-wiseinto the suspension under vigorous stirring to reach pH about 10. Afterbeing aged for 4 h at room temperature, the obtained product was driedat 100 °C for 12 h. The resultant CeO2/TiO2–NTs were finally calcinedat a certain temperature for a specific time according to the Box-Behnken design and designated as CeO2/TiO2–NTs (Fig. 2).

According to initial experiments, four factors, namely hydrothermaltemperature (X1), calcination temperature (X2), hydrothermal time(X3), and CeO2/TiO2 ratio (mol/mol) (X4) were considered to affect thecatalytic properties of CeO2/TiO2–NTs in the degradation of methyleneblue. In the present study, the experimental design of Box–Behnken[41–45] was employed to determine the optimum levels of the vari-ables. The number of experiments (N) required for this design is N =2·k·(k–1) + C0, where k is the factor number and C0 is the replicatenumber of the central point [44]. Thus, a total of 27 runs were per-formed for optimizing these four variables in the current Box–Behnkendesign with 3 runs at the central point. The response of this design wasthe yield of the degradation reaction (Y). The corresponding symbolsand levels are shown in Table 1.

Based on the experimental data, a second-order polynomial modelwas calculated, which provided the relationship between the photo-catalytic degradation yield and the four selected variables. The re-lationship was represented as follows:

∑ ∑ ∑ ∑= + + += = = >

Y X X X Xβ β · β · β · ·i

i ii

ii ii j i

ij i j01

4

1

42

1

3 4

(1)

where Y (%) is the predicted response value; β0 is the intercept term, βiis the linear coefficients; βij is the cross-product coefficients, βii is thequadratic term coefficients, and Xi, Xj are the independent variables.

The design and analysis of this experiment were carried out usingstatistical software MINITAB version 16.

2.3. Photocatalytic degradation experiments

The photocatalytic activity of the synthesised CeO2/TiO2–NTs wasevaluated by studying the degradation of MB in water (100mL, 15mg/L, neutral pH) under the filtrated light of a 250W fluorescent high-pressure mercury lamp (Philip ML 250W) as a visible light source. Thedistance maintained between the lamp and the solution surface was15 cm in all measurements (Fig. 3).

The experiments were conducted under magnetically stirring for30min in the dark, at room temperature to reach the adsorption-des-orption equilibrium; then, the mixture was illuminated with periodicsampling (5mL, 20min). The samples taken from the mixture wereimmediately centrifuged and the filtrates were then analysed by re-cording the absorbance at 664 nm with a UV–vis spectrophotometer(UViline 9400).

The degradation efficiency of MB was calculated according to ex-pression (2)Fig. 1. Molecular formula of methylene blue (MB).

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6000

=−F C C

C·100 %0

0 (2)

where C0 and C are the concentration of the initial and remaining MB,respectively.

The analysis of OH% formation on the sample surface under lightillumination was conducted by means of fluorescence spectroscopyusing terephthalic acid, which readily reacted with OH% to produce ahighly fluorescent product, 2-hydroxyterephthalic acid [46,47]. Briefly,200mg of the obtained CeO2/TiO2–NTs was added to 200mL of a5·10–4 M terephthalic acid solution in 2·10–3 M NaOH, and then thelight illumination of the solution started. Sampling was carried outevery 20min. The solution after filtration was analysed on a fluores-cence spectrophotometer. The product of terephthalic acid hydroxyla-tion – 2-hydroxyterephthalic acid – gave a peak at the wavelength ofabout 425 nm by the excitation with the wavelength of 315 nm.

The chemical oxygen demand (COD) of MB solution was measuredusing the ASTM method [48]. The sample was oxidised by a boilingmixture of chromic and sulfuric acids and refluxed in a strong acidsolution with a known excess of potassium dichromate (K2Cr2O7). Afterdigestion, the remaining unreduced K2Cr2O7 was titrated with ferrousammonium sulfate to determine the amount of K2Cr2O7 consumed, andthe oxidisable matter was calculated in terms of oxygen equivalent. The

samples were analysed in replicates to yield the most reliable data.

2.4. Characterisation techniques

The X-ray powder diffraction (XRD) patterns were collected with aD8 Advance Bruker X-ray diffractometer using CuKα radiation (1.54 Å).The Raman spectra were measured at room temperature using a HoribaXplora Raman spectrophotometer with the excitation laser wavelengthof 532 nm. The transmission electron microscopy (TEM) and scanningelectron microscopy (SEM) images were collected using a JEOL JEM –2100 F (USA) transmission electron microscope and an SEMJMS–5300LV (USA) scanning electron microscope, respectively. Theelemental composition was determined by means of electron dispersiveX-ray (EDX) analysis coupled with HRTEM using JEOL 2100, EDX de-tector with XMax 80 T (Oxford). The atomic absorption measurements(AAS) were made using a Shimadzu AA–7000 flame atomic absorptionspectrometer (Japan). The X-ray photoelectron spectroscopy (XPS)analysis was carried out using a Shimadzu Kratos Axisultra DLD spec-trometer with AlKα as the excitation source. The calibration of thebinding energy and the corrections of the energy shift were accom-plished by assuming that the C1s line lies at 284.6 eV. The diffuse re-flectance spectra (DRS) were measured using a Cary5000 UV–Vis–NIRspectrophotometer, and reflectance spectra were referenced to BaSO4.The nitrogen adsorption-desorption isotherms were performed using aMicromeritics Tristar 3000 device (USA). The Shimadzu RF–5301 PCseries fluorescence spectrometer was used to collect the fluorescenceemission spectra of the samples.

The point of zero charge (pHPZC) of CeO2/TiO2–NTs was estimatedusing the pH drift method [49,50]. Five millilitres of 0.1M NaCl solu-tion and 40mL of distilled water were added to a series of 100mLflasks. The initial pH value (pHi) of the solution was adjusted from 2 to12 by using either 0.1M NaOH or 0.1M HCl solutions. The total volumeof solution in each flask was made exactly to 50mL by adding distilledwater. The 0.01M NaCl solutions with different pH value were ob-tained. Then, 0.1 g of CeO2/TiO2–NTs was added to each flask, and themixtures were stirred for 24 h; the final pH (pHf) of the solutions wasmeasured. The difference between the initial pH and final pH (ΔpH=pHi – pHf) was plotted against pHi. The point of intersection of the curvewith the abscissa, at which ΔpH=0, provided pHPZC.

3. Results and discussion

3.1. Preparation of CeO2/TiO2–NTs using Box–Behnken design

The experimental design matrix of Box–Behnken (B–B) and the re-sults derived from each experiment are represented in Table 2.

According to the polynomial regression model, the correlation ofphotocatalytic degradation yield to corresponding coded values (x1, x2,x3, and x4) of the four variables (hydrothermal temperature, calcinationtemperature, hydrothermal time and CeO2/TiO2 ratio, respectively)was obtained as

Fig. 2. Flow chart of the preparation of CeO2/TiO2–NTs.

Table 1Factors and their levels in the full factorial design.

Factors Low Medium High

Hydrothermal temperature, °C (X1) 140 160 180Calcination temperature, °C (X2) 500 550 600Hydrothermal time, h (X3) 18 20 22CeO2/TiO2 ratio, mol/mol (X4) 0.1 0.3 0.5Coded levels –1 0 1

Fig. 3. Equipment for the photocatalytic experiment.

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6001

= + + − − −

− − − − −

− − − +

Y x x x x x

x x x x x x x

x x x x x x x x

0.8740 0.0249· 0.0246· 0.0250· 0.0502· 0.0755·

0.0770· 0.0774· 0.0012· 0.0010· · 0.0005· ·

0.0008· · 0.0005· · 0.0003· · 0.0005· ·

1 2 3 4 12

22

32

42

1 2 1 3

1 4 2 3 2 4 3 4 (3)

The analysis of variance proved the goodness-of-fit of the models(Table 3). The determination coefficient R2 for the degradation yieldwas calculated as 99.97%, suggesting a good performance for the de-veloped model. R2 adjusted to the degradation yield was 99.94%, i.e.,the developed model for the prediction of the extraction yield mar-ginally differs by± 0.06% from the experimental data. The adequacytest is significant at p-values< 0.05, confirming the adequacy of theselected quadratic model. According to the statistical results with the95% confidence level, each term in the model is significant when its p-value is less than 0.05 (Table 4). Therefore, the reduced form of theregression equation was expressed as follows:

= + + − − −

− −

Y x x x x x

x x

0.8740 0.0249· 0.0246· 0.0250· 0.0502· 0.0755·

0.0770· 0.0774·1 2 3 4 1

2

22

32 (4)

In the regression equation, a negative sign suggested an antagonisticeffect, whereas a positive sign indicated a synergistic effect. From Eq.(4), the similar positive value of the parameter estimate for variables x1and x2 indicated the same level of significance whereas x3 and x4showed the negative relationship.

In this study, the response optimisation by desirability function of

the response surface methodology was employed to find out the optimalparameters to seek for the maximal photocatalytic degradation yield.

The profile for predicted values in MINITAB 16 was employed forthe optimisation process. The optimisation design matrix (Fig. 4) re-presents the maximal photocatalytic degradation (92.9% for MB) at theconditions set: hydrothermal temperature (163 °C), calcination tem-perature (557 °C), hydrothermal time (20 h), and CeO2/TiO2 ratio(0.1 mol·mol–1). The reliability of this prediction was examined by theperformance of five similar experiments at the optimal conditions. Theexperimental degradation yield was 93%, 96%, 94.5%, 95% and 94.2%.The one-sample t-test showed a non-significant difference with re-spective values presented by the model (t (4) = –2.32, p= 0.08).Therefore, the optimal synthesis conditions were used to synthesise theCeO2/TiO2–NTs for further experiments.

3.2. Characterisation of catalysts

The XRD patterns of the as-synthesized TiO2–NTs, TiO2–NTs 550(TiO2–NTs calcinated at 550 °C) and CeO2/TiO2–NTs (prepared at op-timal conditions) are shown in Fig. 5. It can be seen clearly that the as-synthesized TiO2–NTs mainly consist of amorphous phases with lowdiffraction intensity, showing poor crystallinity. However, after beingcalcinated at 550 °C, the characteristic diffractions indexed as anatasephase (JCPDS: 00-021-1272) were observed. In addition, the diffractionpeak around 43°, which can be assigned to the rutile (210) (JCPDS: 00-021-1276) was detected, revealing the remaining of the rutile phase inTiO2–NTs.

The presence of CeO2 in the CeO2/TiO2–NTs sample is based on theappearance of (101) and (200) characteristic peaks (JCPDS: 00-034-0394). The existence of the mixture of anatase and rutile with ceriaindicated the formation of CeO2–TiO2 composite. As shown in Fig. 6aand b, the aggregated CeO2 nanoparticles ranging from 5 to 10 nm insize on the surface of the TiO2 nanotubes could be clearly observed. TheTiO2 nanotubes were hollow and open-ended with an average innerdiameter of 4 nm, an average outer diameter of 10 nm, and about200 nm in length. Ceria dispersed mainly on the surface of TiO2 na-notubes and, therefore, formed a boundary between those particles andthe TiO2 nanotubes. This was confirmed by the measured lattice spacingd of 0.27 nm due to (200) planes of CeO2 and d of 0.35 nm due to (101)of TiO2 obtained from the high-resolution HR–TEM image (Fig. 6c).Furthermore, the EDX spectra of the sample CeO2/TiO2–NTs in Fig. 6dconfirmed the presence of titanium, cerium and oxygen in the sample ofCeO2/TiO2-NTs.

The textural properties of the resulting materials were investigatedby means of nitrogen adsorption/desorption isotherms (Fig. 7). Thecurves of the N2 adsorption-desorption isotherms for all samples weresimilar, in which TiO2–NTs, TiO2–NTs 550, and CeO2/TiO2–NTs

Table 2Experimental design in coded units and response.

Exp. x1 x2 x3 x4 Yield / %

1 0 −1 0 1 0.7222 −1 0 1 0 0.6703 0 0 −1 1 0.7684 1 0 1 0 0.7215 0 −1 0 −1 0.8196 −1 0 0 1 0.7237 0 0 0 0 0.8748 0 1 0 1 0.7729 0 1 0 −1 0.87010 1 0 0 −1 0.87211 −1 0 −1 0 0.72012 1 1 0 0 0.76813 0 0 1 −1 0.82114 0 1 1 0 0.71915 1 0 −1 0 0.77316 0 0 0 0 0.87417 0 0 0 0 0.87318 0 1 −1 0 0.76919 −1 0 0 −1 0.82220 0 −1 1 0 0.67021 0 −1 −1 0 0.71822 −1 1 0 0 0.72123 0 0 1 1 0.71924 1 −1 0 0 0.72325 0 0 −1 −1 0.87226 −1 −1 0 0 0.67227 1 0 0 1 0.770

Table 3ANOVA for the fit of the experimental data to the response surface.

Source Degree offreedom

Adjusted sum ofsquares

Adjusted meansquares

p-value

Regression 14 0.122255 0.008732 0.000Linear 4 0.052403 0.013101 0.000Square 4 0.069843 0.017461 0.000Interaction 6 0.000009 0.000002 0.746Residual error 12 0.000033 0.000003Lack-of-fit 10 0.000033 0.000003 0.096Pure error 2 0.000001 0.000000Total 26 0.122288

Table 4ANOVA for the degradation yield.

Source Regression coefficient F-value p-value

x1 0.0249 2688.75 0.00x2 0.0246 2617.29 0.00x3 –0.0250 2706.77 0.00x4 –0.0502 10899.37 0.00x12 –0.0755 10984.02 0.00x22 –0.0770 11424.56 0.00x32 –0.0774 11536.05 0.00x42 –0.0012 2.62 0.13x1·x2 –0.0010 1.44 0.25x1·x3 –0.0005 0.36 0.56x1·x4 –0.0008 0.81 0.39x2·x3 –0.0005 0.36 0.56x2·x4 –0.0003 0.09 0.77x3·x4 0.0005 0.36 0.56Model 7.22 0.001

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6002

exhibited the type IV isotherm according to IUPAC classification, in-dicating the mesoporous structure of the materials, and the type H3hysteresis loop suggesting the presence of slit-shaped pores. The syn-thesised TiO2–NTs had a much higher BET surface area (247m2 g–1)than both TiO2–NTs 550 (64m2 g–1) and CeO2/TiO2–NTs (66m2 g–1).

The XPS spectra were used to determine the elemental chemicalstate on the surface of the synthesised CeO2/TiO2–NTs, and the XPSanalysis of Ce 3d, Ti 2d and O 1 s is demonstrated in Fig. 8. The Ce 3dspectrum (Fig. 8a) was relatively complex, and the identification of thechemical states of Ce 3d was labelled according to the convention es-tablished by Burroughs et al. [51], where v and u indicate the spin-orbitcoupling states of 3d5/2 and 3d3/2, respectively. The binding energy ofCe 3d5/2 at 881, 885.5, 887.8, and 897.6 eV corresponded with v0, vꞌ, vꞌꞌand vꞌꞌꞌ, while the binding energy of Ce 3d3/2 at 899.4, 902, 907.2 and916.1 eV was respective for u0, uꞌ, uꞌꞌ and uꞌꞌꞌ. The peaks at v0, vꞌ, u0 anduꞌ were characteristic of Ce3+, where uꞌ/vꞌ was related to the Ce(3d94f1) O (2p6) final state, and u0/v0 was assigned to the Ce (3d94f2) O(2p5) state [52,53]. The peaks labelled as vꞌꞌ, vꞌꞌꞌ, uꞌꞌ, and uꞌꞌꞌ were at-tributed to Ce4+ in CeO2 [40,52,54]. uꞌꞌꞌ/vꞌꞌꞌ was ascribed to the pri-mary photoemission from Ce4+–O2 [55], corresponding with the Ce(3d94f°) O (2p6) state. The uꞌꞌ/vꞌꞌ doublet was resulting from thetransfer of two electrons from the O 2p orbital to an empty Ce 4f orbitalwith the (3d94f1) O (2p5) final state. The transfer of electrons from theO 2p to Ce 4f orbitals in the photoemission would increase the electron

density of Ce4+, thus decreasing the attraction of the Ce nucleus to-wards the electrons on the outermost layer while enhancing the re-pulsion between electrons. This is also why the more electrons weretransferred from the O 2p orbitals, the poorer the binding energy of theCe4+ 3d core level would become [56]. Based on these observations, itcan be concluded that a mixture of Ce4+/Ce3+ oxidation states existedon the surface of the synthesised CeO2/TiO2–NTs catalyst. The resultswere well in line with previous photoemission data reported by severalauthors on CeO2/TiO2 nanotube materials [2,40,57,36,58], as well asmixed CeO2/TiO2 materials obtained via other synthesis methods withdifferent morphologies [25,53,56,59,60]. Concretely, Graciani et al.have used density functional calculations to demonstrate that Ce3+ ionsare strongly stabilised due to the energy decrease of the Ce 4f levels as aresult of mixing with the O 2p band of titania at the ceria-titania in-terface [61]. It is clear that the vehement interaction between CeO2 andTiO2 promoted the reduction of Ce4+ to Ce3+ and led to the presence ofCe3+ in the CeO2/TiO2 materials. Fig. 8b shows the Ti 2p core-levelspectrum of CeO2/TiO2–NTs. The binding energies of Ti 2p3/2 and 2p1/2in CeO2/TiO2–NTs were 458.4 and 464.2 eV, corresponding with ty-pical characteristics of the octahedral coordinated Ti4+ ions [62]. Thepresence of CeO2 maintained the binding energies of Ti 2p, and Ti ex-isted as Ti4+ in the doped materials, which is in accordance with thepreviously reported results [25,56]. The O 1 s spectra (Fig. 8c) emergedwidely and unsymmetrically, which indicates that at least three types ofoxygen constituents were present at the material surface. The two peakslocated at 528.8 and 530.0 eV could be ascribed to the O 1 s electronbinding energy of oxygen (O2–) in anatase TiO2 and/or CeO2. And thefitting data at 532.2 eV should be assigned to the hydroxyl groups in thematerial [63–65].

The Raman spectra were employed to define the lattice structure ofthe CeO2–TiO2 composite (Fig. 9). The result outputs asserted the ex-istence of anatase in both samples synthesised, via the presence ofcharacteristic signals for the tetragonal phase of anatase around142 cm–1; 395 cm–1, 515 cm−1 and 637 cm–1 [66–71]. The shoulder at195 cm–1 was also contributed to anatase [72]. The CeO2 /TiO2–NTssample provided a strong band centred at 464 cm–1, which was attrib-uted to the typical vibrational mode of the cubic fluorite lattice[54,60,73]. The observed band at 241 cm–1was assigned to the surfacemode of the CeO2(111) surface, and in a good agreement with theRaman spectra of ceria (CeO2) powders with varying crystal sizes [74].In addition, an increased peak at 195 cm–1 was attributed to the latticemode (Na–O–Ti peaks) according to Marques et al. [75]. This resultagain confirms that CeO2 was associated with titania on the anatasestructure.

Fig. 4. Optimisation plots for the photocatalytic degradation yield of MB over CeO2/TiO2–NTs.

Fig. 5. XRD patterns of TiO2–NTs; TiO2–NTs 550; CeO2/TiO2–NTs.

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6003

In order to investigate the electronic states of the catalysts, theUV–vis diffuse reflectance spectroscopy (DRS) studies were performedon TiO2–NTs 550, CeO2 and CeO2/TiO2–NTs. Compared with bareTiO2-TiO2 and bare CeO2, CeO2/TiO2–NTs clearly exhibited broaderabsorption in the visible region (λ=400–600 nm), displaying a slightred shift in the optical adsorption (Fig. 10a). The band gap (Eg) of theas-prepared samples could also be calculated by using Tauc’s expres-sion: = −α h υ A h υ E· · ·( · )g

12 , (where α is the absorption coefficient, A is

the parameter that is related to the effective mass associated with the

valence and conduction bands; h is the Planck constant; υ is the fre-quency [76]. The band gap values were calculated by plotting α h υ( · · )

12

versus h υ· . The band gap of CeO2/TiO2–NTs was found to be 2.64 eV,lower than that of TiO2–NTs (3.08 eV) and CeO2 (2.93 eV) (Fig. 10b).This red shift of the adsorption edge indicated the enhanced ability ofthe CeO2/TiO2–NTs hybrid catalyst to absorb visible light. The red shiftin the optical transition, as well as the decreased band gap value of theCeO2/TiO2–NTs composite, was suggested to be relative to the presenceof the Ce3+ ions with one electron in the strong localised 4f orbitals asobserved from the XPS spectra. The existence of the Ce3+ ions in thesynthesised CeO2/TiO2–NTs with respect to the 4f occupation (4f1)resulted in an additional energy state, leading to the visible light ab-sorption via the electron transfer transition from 4f to 5d orbitals, andthus reducing the band gap [56,65,77].

The band edge energy at the interface of the n-semiconductor wascalculated according to the equations proposed by Xu and Schoonen asfollows [78]:

=E χ E E– –0.5·CB e g (5)

=E E E–VB g CB (6)

where ECB is the conduction band edge energy; EVB is the valence bandedge energy; χ is the electronegativity (χ=5.56 eV for CeO2 andχ=5.81 eV for TiO2 [79]); Ee is the free energy of electrons with re-spect to the normal hydrogen electrode (NHE) (Ee=4,5 eV [79]); Eg isthe band gap of the semiconductor.

The values of ECB and EVB for TiO2 calculated from Eqs. (5) and (6)were –0.23 and 2.85 eV, respectively; and those for CeO2 were –0.405and 2.525 eV, respectively. The position of energy levels of CeO2 andTiO2 is illustrated in Fig. 11. It can be seen from the figure that thepromoted photocatalytic activity of CeO2/TiO2-NTs is expected for tworeasons: (i) the strong light adsorption of CeO2 in the visible region, (ii)

Fig. 6. TEM images of (a) TiO2–NTs, (b) CeO2/TiO2–NTs, (c) HRTEM image of CeO2/TiO2–NTs and (d) EDX spectrum of CeO2/TiO2–NTs.

Fig. 7. Nitrogen adsorption/desorption isotherms of TiO2–NTs, TiO2–NTs 550,and CeO2/TiO2–NTs.

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6004

the transfer of electrons from ECB of CeO2 to ECB of TiO2 has beenconsidered to prevent the fast recombination of photoexcited e– and h+.

3.3. Photocatalytic activity

3.3.1. Photocatalytic behaviours of CeO2/TiO2–NTs3.3.1.1. Comparison of several catalysts. The adsorption kinetics andphotocatalytic kinetics over several catalysts (TiO2–NTs 550, CeO2,CeO2/TiO2–NTs, and commercial P25 for the sake of comparison) are

represented in Fig. 12. The degradation efficiency of MB was close tozero after light illumination for 150min without a catalyst (blanksample). This means that the photolysis of MB was negligible. The MBadsorption over CeO2 and TiO2-NTs was around 5% and 90% underdark adsorption. Subsequently, the degradation efficiency of MBremained constant around these values when visible light was appliedfor a further 120min, indicating that CeO2 or TiO2-NTs could notcatalyse the photodegradation of MB under these conditions.Meanwhile, CeO2/TiO2–NTs and P25 exhibited a different behaviourtowards the MB solution. After dark adsorption when the discolouringof MB was 68% for the former and 15% for the latter, these twomaterials exhibited catalytic activity under light illumination. CeO2/TiO2–NTs provided a photodegradation efficiency of MB at 94.6% after60min of irradiation and practically 100% after 120min, while P25produced a degradation efficiency of only 68% after 120min ofillumination. These results indicated that the composite of TiO2 andCeO2 successfully improved the photodegradation capability of TiO2

nanotubes.The UV–Vis spectra for the photocatalytic degradation of MB over

CeO2/TiO2–NTs showed that the maximum absorption at 664 nm(electron transfer π–π* in the MB structure) decreased with the increasein light illumination time (Fig. 13a). To confirm the mineralisation ofMB over CeO2/TiO2–NTs catalyst, the change of COD of reaction pro-ducts with time was analysed (Fig. 13b). The initial COD was28.2 mg L–1, and its decrease became faster as the illumination timeincreased, reaching 10.68mg L–1 after 120min. These results confirmedthe effectiveness of CeO2/TiO2–NTs as a photocatalyst for MB de-gradation under visible light. After 120min of illumination, the total

Fig. 8. XPS spectra of a) Ce 3d; b)Ti 2p, and c)O 1 s for CeO2/TiO2–NTs.

Fig. 9. Raman spectra of TiO2–NTs 550 and CeO2/TiO2–NTs.

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6005

decolourisation of MB practically occurred, while around 37.9% of CODreduction was obtained. The difference between the degradation andmineralisation could be attributed to the existence of intermediateproducts.

3.3.1.2. Leaching experiment. The contribution of the active metal ionsleached from the solid catalyst in the total catalytic activity is animportant problem in the application of heterogeneous catalysts forliquid phase processes [80]. The leaching experiment was carried out inthe way that CeO2/TiO2–NTs catalyst was filtered after 70min ofreaction (Fig. 14). The MB conversion stopped at an 85% efficiencydespite the fact that light continued to illuminate a further 80min, andTi and Ce were also absent in the solution (AAS analysis). This indicatesthat the catalyst was stable under working conditions. CeO2/TiO2–NTswas a true heterogeneous catalyst in the MB photocatalyticdegradation.

3.3.1.3. Effect of pH. pH of the solution is one of the important factorsaffecting the efficiency of decomposition of organic matter duringphotochemical catalysis [81]. This decomposition is significantlydependent on the number of the adsorbed dye cations and the OHradicals on the surface of the catalyst, and this, in turn, depends on thestate of the catalyst surface.

The MB degradation efficiency increased sharply when pH increased

Fig. 10. a) UV–Vis diffuse reflectance spectra and b) Tauc’ plots of TiO2–NTs, CeO2, and CeO2/TiO2–NTs.

Fig. 11. Energy of valence band edge and conduction band edge for TiO2 and CeO2.

Fig. 12. Degradation efficiency of MB solution during visible light illuminationover TiO2–NTs, P25, CeO2, CeO2/TiO2–NTs and blank (V=100mL,C0(MB) = 15 ppm, mcatalyst =0.08 g, illumination time: 120min, room tem-perature).

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6006

from 3 to 4, followed by a slow increase when the pH changed from 4 to8 but then decreased significantly when the pH increased to 12(Fig. 15). The point of zero charge (pHPZC) of CeO2/TiO2–NTs estimatedby the pH drift method was approximately 4 (inset of Fig. 15). At pH< 4, the MB molecule is neutral (pKa=3.8) [82], and the surface ofthe CeO2/TiO2–NTs was charged positively due to protonation. There-fore, the van der Waals interaction between MB and CeO2/TiO2–NTscould be dominant.

Such a poor interaction results in a very low efficiency of the pho-tochemical degradation reaction. Furthermore, with the positivelycharged surface, the hydroxyl ion supply required for the free radicalformation is limited, which is important for photocatalytic decom-position. The MB degradation increases with increasing pH because theincreasing electrostatic interaction between the negatively chargedsurface and positive cationic dye causes a stronger photocatalytic re-action. The higher the pH is, the higher the number of hydroxyl ions atthe surface of CeO2/TiO2–NTs will be, and hydroxyl ions give hydroxylradicals in the following equation [46,81]:

+ →+h OH OH•– (7)

However, the photochemical degradation is inhibited when pH istoo high because the hydroxyl ion may compete with the MB moleculein the adsorption on the photocatalyst surface.

3.3.1.4. Formation of free radicals. The free radical generation by

photoinduced electron-hole pairs was confirmed by the fluorescenceemission spectrum. Fig. 16a shows the induction of fluorescence from5×10−4 M terephthalic acid in the 2×10−3 M NaOH solution. Theincrease in the fluorescence intensity against illumination time at425 nm was observed. The fluorescence intensity by UV lightillumination in the terephthalic acid solutions increased almostlinearly with time. Consequently, we can conclude that the formationof OH. at the CeO2/TiO2–NTs interface was proportional to the lightillumination.

In order to confirm the formation of hydroxyl radicals on the cat-alytic interface by visible light illumination, tert-butanol was used asthe free radical scavenger. In the presence of tert-butanol, a markedreduction in the MB photochemical catalytic activity was observed, andthe decomposition efficiency decreased sharply when increasing theamount of tert-butanol (Fig. 16b). It can be seen from the figure that thedegradation efficiency of MB decreased from 97% to 64.6% and 58.3%as 0.1 mL and 0.2mL tert-butanol was added, respectively. These resultsconfirm that the free radical mechanism plays an important role for thephotocatalytic degradation.

3.3.1.5. Mechanism of photocatalytic reactions over CeO2/TiO2–NTs. Upon light illumination, at the hetero-junction interface ofCeO2 and TiO2, first, CeO2 absorbed light and the electron was excitedto move from the valence band (VB) to the conduction band (CB). Then,the photoexcited electrons jumped to the conduction band of TiO2

where the CB level (–0.405 eV) of CeO2 is more negative than that ofTiO2 (–0.23 eV). These electrons easily reduced the surface oxygen toproduce a large number of reactive oxy radicals such as superoxideradical ion −O •2 and hydroxyl radicals OOH%/OH% on the surface ofCeO2/TiO2–NTs. These oxyradicals reacted with MB giving CO2/H2Othrough a number of intermediates. At the same time, thephotoexcitation provided photoinduced holes at the valence band ofCeO2, which acted as a strong oxidant to oxidise H2O or HO– [60,64],and produced the active species such as OH. (Fig. 11). Therefore, alarge number of oxygenated species on the surface of the catalyst mightsignificantly initiate the photodegradation of MB into less harmfulminerals under light illumination. In addition, the deviation from idealstoichiometry, i.e. CeO2−x (Ce3+ and Ce4+), leads to an occupation ofthe f band, which splits into an occupied part and an unoccupied part[83,84]. Therefore, the activity was found to be directly related to thepresence of the Ce3+ species which caused the extended absorption inthe visible-light region due to the populated 4f states at the interfacebetween ceria and titania. The formation of the inter-band transitionfrom Ce3+ facilitated the interfacial charge transfer, thus improving thephotocatalytic degradation efficiency of the CeO2–TiO2 composites.Overall, the enhanced photocatalytic activity of the hybrid catalysts

Fig. 13. Degradation of MB over CeO2/TiO2–NTs: a) UV–Vis spectra; b) COD value (V=100mL, C0(MB) = 15 ppm, mcatalyst =0.08 g, illumination time: 120min,room temperature).

Fig. 14. Leaching experiments of CeO2/TiO2–NTs (V=100mL,C0(MB) = 15 ppm,mcatalyst =0.08 g, illumination time: 120min, room tem-perature).

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6007

involving CeO2 and TiO2 could be attributed to the synergistic effectsbetween two oxides due to the presence of hetero-junctions and theCe3+ species in the composites.

The degradation reactions of MB over CeO2/TiO2–NTs could be il-lustrated as follows [65,85,86],

CeO2/TiO2 + hλ → h+ + e– (8)

h+ + H2O → H+ + OH% (9)

h+ + OH– → OH% (10)

+ → −e O O •–2 2 (11)

+ → +−O H O OOH OH• •2 2– (12)

2 OOH% → H2O2 + O2 (13)

H2O2 + e– → OH– + OH% (14)

Ce4+ + e– → Ce3+ (15)

+ → ++ + −Ce O Ce O •32

42 (16)

+ →−OH O MB products( •, •)2 (17)

3.3.2. Photocatalytic kinetics of MB degradation over CeO2/TiO2–NTsThe kinetics of adsorption and photocatalytic degradation of MB on

CeO2/TiO2–NTs are presented in Fig. 17.The dark adsorption took place quickly and reached equilibrium at

around 30min. Afterwards, the MB concentration continued to de-crease due to the reaction under visible light. It is supposed that MB wasadsorbed on the catalyst and then it was decomposed on the catalystsurface under visible illumination according to the Langmuir-Hinshelwood model.

The overall reaction could be illustrated as follows

+ ⇄ … ⟶−

X MB X MB products( )*k

k k

1

1 2

where MB refers to methylene blue; X is the catalyst (CeO2/TiO2–NTs);

Fig. 15. Effect of pH on the degradation yield (V=100mL, C0(MB) = 15 ppm, mcatalyst =0.08 g, illumination time: 120min, room temperature, pH: 3–12); the insetpresents the pHPZC estimated using pH drift method.

Fig. 16. a) Fluorescence spectra observed during illumination of CeO2/TiO2–NTs in 2× 10−3 M NaOH solution of 5× 10–4 M terephthalic acid. (V=200mL;mcatalyst =200mg, illumination time: 120min, room temperature); b. Influence of tert-butanol on the degradation of MB (V=100mL; mcatalyst =0.08 g, illumi-nation time: 120min, room temperature).

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6008

k1 is the forward adsorption rate constant; k–1 is the backward ad-sorption rate constant; k2 is the photocatalytic rate constant.

Assuming that the photocatalytic degradation is the slow rate-de-termining step, the rate of degradation is expressed as

= − =r dCdt

k θ·r (18)

where θ is the fraction of the surface covered by MB.At equilibrium, θ is expressed as follows

=+

θ K CK C·

1 ·L

L (19)

Therefore,

= − =+

r dCdt

k K CK C

· ·1 ·r

L

L (20)

It is known that when the initial concentrations of the reactants are low,then KL·C< <1, and the photocatalytic reactions follow the pseudo-first-order kinetics model [87]. The kinetics Eq. (20) can be written as

− =dCdt

k C·app (21)

where kapp (= kr·KL) is the rate constant (min−1).Integrating Eq. (21) (under the boundary conditions C=C0a at

t=0; C0a is the initial concentration of MB after dark adsorption, and Crepresents the concentration of MB at the reaction time t) provided Eq.(22)

=CC

k tln ·aapp

0(22)

The linear plot of C Cln( )a0 vs. t yields kapp. High coefficients ofregression, R2 (0.911–0.989) confirmed that the MB photocatalytic

degradation fitted the L–H first-order kinetic model well. Furthermore,as shown in Table 5, the apparent first-order rate constants kapp de-clined with the increase of initial MB concentrations. This might be theresult of the generated intermediate products during the photocatalyticreaction; specifically, the MB photodegradation productivity becamepoorer since a great number of intermediates were adsorbed on thesurface of CeO2–TiO2–NTs, which slowed down the overall reactionrate.

The initial concentration of MB also affected its initial rate of pho-todegradation. As seen from Table 5, the initial photodegradation rateincreased with the MB concentration up to 15 ppm from 0.18mgL–1 min–1 to 0.48mg L–1 min–1; however, it decreased dramatically to0.13mg L–1 min–1 when the original MB concentration was 35 ppm.

The initial rate could be expressed as follows

=+

K C k K CK C

· · ·1 ·app a r

L a

L a0

0

0 (23)

Rearranging Eq. (23) gives

= +k

Ck k K

1 1·app

oa

r r L (24)

The plot of k1 app versus C0a showed a linear variation with a highcoefficient of regression (R2= 0.969, p= 6.10–5). The values of kr andKL calculated from the slope and the intercept of the straight line for thephotocatalytic process were 0.103mg L–1 min–1 and 0.840 L·mg–1, re-spectively.

3.3.3. Cycle photocatalytic degradation studiesThe stability of the catalyst was evaluated by using it repeatedly.

During the first two cycles, the used catalyst was separated by cen-trifuging, then washed with deionised water, and dried at 100 °C. Thedecrease of MB concentration in each cycle is represented in Fig. 18.After each cycle, a slight activity loss was observed (around 2%). Thisindicates that the CeO2/TiO2–NTs was a robust stable catalyst.

4. Conclusions

The synthesis and photocatalytic activity of CeO2/TiO2–NTs in thedegradation of methylene blue were detailed in this study. TheBox–Behnken design of the response surface methodology was em-ployed to optimize the photocatalytic degradation of MB on the syn-thesised CeO2/TiO2–NTs. The model prediction of RSM fitted well withthe experimental data with R2 and adjusted R2 of 89.4% and 77.0%,respectively. The optimization results showed that the maximal re-moval yield (93.4%) was obtained at the optimum synthesis conditions:

Fig. 17. Kinetics of adsorption and photocatalytic degradation of MB overCeO2/TiO2–NTs at different initial MB concentrations (V=100mL, C0

(MB) = 5–35 ppm, mcatalyst =0.08 g, irradiation time: 120min, room tempera-ture).

Table 5Kinetic parameters for the degradation of MB.

C0 (ppm) kapp (min–1) r0 (mg L–1 min–1) R2 p-value

5 0.036 0.18 0.989 < 0.00110 0.034 0.34 0.979 < 0.00115 0.032 0.48 0.970 < 0.00120 0.014 0.28 0.967 < 0.00125 0.007 0.15 0.911 < 0.00130 0.005 0.15 0.958 < 0.00135 0.004 0.13 0.956 < 0.001

Fig. 18. Regeneration of CeO2/TiO2–NTs after three recycles (V=100mL, C0

(MB) = 15 ppm, mcatalyst =0.08 g, illumination time: 120min, room tempera-ture).

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6009

hydrothermal temperature of 160 °C, calcination temperature of 550 °C,hydrothermal time of 20 h, and CeO2/TiO2 molar ratio of 0.1. The re-sults clearly demonstrated that the response surface methodology(RSM) with the Box–Behnken design was one of the reliable methodsfor modeling and optimisation of the synthesis variables. CeO2 as adopant for TiO2–NTs brought a red shift and drove the band gap to thevisible light region and efficiently hindered the recombination of pho-toinduced electron-hole pairs. Consequently, the photocatalytic activityof CeO2/TiO2 increased significantly in the visible region comparedwith each individual oxide. The synthesised catalyst was stable afterthree recycles without metal leaching. These results suggested CeO2/TiO2–NTs to be a promising catalyst for the heterogeneous photo-catalytic dye degradation in the visible region.

Acknowledgements

This research is funded by National Foundation for Science andTechnology Development (NAFOSTED), Vietnam; grant IDs: 104.06-2017.311.

References

[1] H. Eskandarloo, A. Badiei, M.A. Behnajady, TiO2/CeO2 hybrid photocatalyst withenhanced photocatalytic activity: optimization of synthesis variables, Ind. Eng.Chem. Res. 53 (2014) 7847–7855.

[2] W. Xue, G. Zhang, X. Xu, X. Yang, C. Liu, Y. Xu, Preparation of titania nanotubesdoped with cerium and their photocatalytic activity for glyphosate, Chem. Eng. J.167 (2011) 397–402.

[3] M. Hussain, R. Ceccarelli, D.L. Marchisio, D. Fino, N. Russo, F. Geobaldo, Synthesis,characterization, and photocatalytic application of novel TiO2 nanoparticles, Chem.Eng. J. 157 (2010) 45–51.

[4] K. Shankar, J.I. Basham, N.K. Allam, O.K. Varghese, G.K. Mor, X. Feng, M. Paulose,J.A. Seabold, K.-S. Choi, C.A. Grimes, Recent advances in the use of TiO2 nanotubeand nanowire arrays for oxidative photoelectrochemistry, J. Phys. Chem. C 113(2009) 6327–6359.

[5] S. Pavasupree, S. Ngamsinlapasathian, Y. Suzuki, S. Yoshikawa, Synthesis and dye-sensitized solar cell performance of nanorods/nanoparticles TiO2 from high surfacearea nanosheet TiO2, J. Nanosci. Nanotechnol. 6 (2006) 3685–3692.

[6] F. Zhang, C. Zhang, H. Peng, H. Cong, H. Qian, Near‐infrared photocatalytic up-conversion nanoparticles/TiO2 nanofibers assembled in large scale by electro-spinning, Part. Part. Syst. Charact. 33 (2016) 248–253.

[7] W. Guo, F. Zhang, C. Lin, Z.L. Wang, Direct growth of TiO2 nanosheet arrays oncarbon fibers for highly efficient photocatalytic degradation of methyl orange, Adv.Mater. 24 (2012) 4761–4764.

[8] B.-K. Min, S.-D. Choi, SnO2 thin film gas sensor fabricated by ion beam deposition,Sensors Actuators B Chem. 98 (2004) 239–246.

[9] Q. Tian, W. Wu, S. Yang, J. Liu, W. Yao, F. Ren, C. Jiang, Zinc oxide coating effectfor the dye removal and photocatalytic mechanisms of flower-like MoS 2 nano-particles, Nanoscale Res. Lett. 12 (2017) 221.

[10] Q. Tian, W. Yao, Z. Wu, J. Liu, L. Liu, W. Wu, C. Jiang, Full-spectrum-activated Z-scheme photocatalysts based on NaYF 4: Yb 3+/Er 3+, TiO 2 and Ag 6 Si 2 O 7, J.Mater. Chem. A 5 (2017) 23566–23576.

[11] D.V. Bavykin, V.N. Parmon, A.A. Lapkin, F.C. Walsh, The effect of hydrothermalconditions on the mesoporous structure of TiO2 nanotubes, J. Mater. Chem. 14(2004) 3370.

[12] F. Song, Y. Zhao, Q. Zhong, Adsorption of carbon dioxide on amine-modified TiO2nanotubes, J. Environ. Sci. 25 (2013) 554–560.

[13] K. Rajeshwar, M.E. Osugi, W. Chanmanee, C.R. Chenthamarakshan, M.V.B. Zanoni,P. Kajitvichyanukul, R. Krishnan-Ayer, Heterogeneous photocatalytic treatment oforganic dyes in air and aqueous media, J. Photochem. Photobiol. C Photochem.Rev. 9 (2008) 171–192.

[14] M. Kim, S.-H. Hwang, S.K. Lim, S. Kim, Effects of ion exchange and calcinations onthe structure and photocatalytic activity of hydrothermally prepared titanate na-notubes, Cryst. Res. Technol. 47 (2012) 1190–1194.

[15] V. Likodimos, T. Stergiopoulos, P. Falaras, J. Kunze, P. Schmuki, Phase composi-tion, size, orientation, and antenna effects of self-assembled anodized titania na-notube arrays: a polarized micro-raman investigation, J. Phys. Chem. C 112 (2008)12687–12696.

[16] G.K. Mor, O.K. Varghese, M. Paulose, K. Shankar, C.A. Grimes, A review on highlyordered, vertically oriented TiO2 nanotube arrays: fabrication, material properties,and solar energy applications, Sol. Energy Mater. Sol. Cells 90 (2006) 2011–2075.

[17] C.L. Wong, Y.N. Tan, A.R. Mohamed, A review on the formation of titania nanotubephotocatalysts by hydrothermal treatment, J. Environ. Manage. 92 (2011)1669–1680.

[18] T. Kasuga, M. Hiramatsu, A. Hoson, T. Sekino, K. Niihara, Titania nanotubes pre-pared by chemical processing, Adv. Mater. 11 (1999) 1307.

[19] J. Yu, H. Yu, Facile synthesis and characterization of novel nanocomposites of ti-tanate nanotubes and rutile nanocrystals, Mater. Chem. Phys. 100 (2006) 507–512.

[20] X. Chen, S. Cao, X. Weng, H. Wang, Z. Wu, Effects of morphology and structure of

titanate supports on the performance of ceria in selective catalytic reduction of NO,Catal. Commun. 26 (2012) 178–182.

[21] A.A. Farghali, A.H. Zaki, M.H. Khedr, B. Suef, Hydrothermally synthesized TiO2nanotubes and nanosheets for photocatalytic degradation of color yellow sunset,Int. J. Adv. Res. 2 (2014) 285–291.

[22] L. Peng, T. Xie, Y. Lu, H. Fan, D. Wang, Synthesis, photoelectric properties andphotocatalytic activity of the Fe2O3/TiO2 heterogeneous photocatalysts, Phys.Chem. Chem. Phys. 12 (2010) 8033.

[23] R. Nadarajan, W.A. Wan Abu Bakar, R. Ali, R. Ismail, Effect of structural defectstowards the performance of TiO2/SnO2/WO3 photocatalyst in the degradation of1,2-dichlorobenzene, J. Taiwan Inst. Chem. Eng. 64 (2016) 106–115.

[24] J. Wang, G. Ji, Y. Liu, M.A. Gondal, X. Chang, Cu2O/TiO2 heterostructure nanotubearrays prepared by an electrodeposition method exhibiting enhanced photocatalyticactivity for CO2 reduction to methanol, Catal. Commun. 46 (2014) 17–21.

[25] Y. Wang, B. Li, C. Zhang, L. Cui, S. Kang, X. Li, L. Zhou, Ordered mesoporous CeO2-TiO2 composites: highly efficient photocatalysts for the reduction of CO2 with H2Ounder simulated solar irradiation, Appl. Catal. B Environ. 130–131 (2013) 277–284.

[26] M.M. Momeni, Y. Ghayeb, M. Davarzadeh, Single-step electrochemical anodizationfor synthesis of hierarchical WO3-TiO2 nanotube arrays on titanium foil as a goodphotoanode for water splitting with visible light, J. Electroanal. Chem. Lausanne(Lausanne) 739 (2015) 149–155.

[27] J. Qian, Z. Chen, C. Liu, X. Lu, F. Wang, M. Wang, Improved visible-light-drivenphotocatalytic activity of CeO2 microspheres obtained by using lotus flower pollenas biotemplate, Mater. Sci. Semicond. Process. 25 (2014) 27–33.

[28] T. Montini, M. Melchionna, M. Monai, P. Fornasiero, Fundamentals and catalyticapplications of CeO2-based materials, Chem. Rev. 116 (2016) 5987–6041.

[29] D.A. Andersson, S.I. Simak, N.V. Skorodumova, I.A. Abrikosov, B. Johansson,Theoretical study of Ce O2 doped with tetravalent ions, Phys. Rev. B - Condens.Matter Mater. Phys. 76 (2007) 1–10.

[30] C. Sun, H. Li, L. Chen, Nanostructured ceria-based materials: synthesis, properties,and applications, Energy Environ. Sci. 5 (2012) 8475.

[31] M.R. Mohammadi, D.J. Fray, Nanostructured TiO2-CeO2 mixed oxides by an aqu-eous sol-gel process: effect of Ce:Ti molar ratio on physical and sensing properties,Sens. Actuators B Chem. 150 (2010) 631–640.

[32] M. Reli, N. Ambrožová, M. Šihor, L. Matějová, L. Čapek, L. Obalová, Z. Matěj,A. Kotarba, K. Kočí, Novel cerium doped titania catalysts for photocatalytic de-composition of ammonia, Appl. Catal. B Environ. 178 (2015) 108–116.

[33] Q. Huang, T. Gao, F. Niu, D. Chen, Z. Chen, L. Qin, X. Sun, Y. Huang, K. Shu,Preparation and enhanced visible-light driven photocatalytic properties of Au-loaded TiO2 nanotube arrays, Superlattices Microstruct. 75 (2014) 890–900.

[34] C. Fan, P. Xue, Y. Sun, Preparation of Nano-TiO2 doped with cerium and its pho-tocatalytic activity, J. Rare Earths 24 (2006) 309–313.

[35] T. Lundstedt, E. Seifert, L. Abramo, B. Thelin, Å. Nyström, J. Pettersen, R. Bergman,Experimental design and optimization, Chemometr. Intell. Lab. Syst. 42 (1998)3–40.

[36] N.A. Eleburuike, W.A. Wan Abu Bakar, R. Ali, M.F. Omar, Photocatalytic de-gradation of paraquat dichloride over CeO 2 -modified TiO 2 nanotubes and theoptimization of parameters by response surface methodology, RSC Adv. 6 (2016)104082–104093.

[37] C.-C. Tsai, H. Teng, Regulation of the physical characteristics of titania nanotubeaggregates synthesized from hydrothermal treatment, Chem. Mater. 16 (2004)4352–4358.

[38] N. Viriya-empikul, T. Charinpanitkul, N. Sano, A. Soottitantawat, T. Kikuchi,K. Faungnawakij, W. Tanthapanichakoon, Effect of preparation variables on mor-phology and anatase–brookite phase transition in sonication assisted hydrothermalreaction for synthesis of titanate nanostructures, Mater. Chem. Phys. 118 (2009)254–258.

[39] N. Liu, X. Chen, J. Zhang, J.W. Schwank, A review on TiO2-based nanotubes syn-thesized via hydrothermal method: formation mechanism, structure modification,and photocatalytic applications, Catal. Today 225 (2014) 34–51.

[40] H. Zhao, Y. Dong, P. Jiang, G. Wang, J. Zhang, Highly dispersed CeO 2 on TiO 2nanotube: a synergistic nanocomposite with superior peroxidase-like activity, ACSAppl. Mater. Interfaces 7 (2015) 6451–6461.

[41] S.L.C. Ferreira, R.E. Bruns, H.S. Ferreira, G.D. Matos, J.M. David, G.C. Brandão,E.G.P. da Silva, L.A. Portugal, P.S. dos Reis, A.S. Souza, W.N.L. dos Santos, Box-Behnken design: an alternative for the optimization of analytical methods, Anal.Chim. Acta 597 (2007) 179–186.

[42] S.L.C. Ferreira, R.E. Bruns, E.G.P. da Silva, W.N.L. dos Santos, C.M. Quintella,J.M. David, J.B. de Andrade, M.C. Breitkreitz, I.C.S.F. Jardim, B.B. Neto, Statisticaldesigns and response surface techniques for the optimization of chromatographicsystems, J. Chromatogr. A 1158 (2007) 2–14.

[43] M.A. Bezerra, R.E. Santelli, E.P. Oliveira, L.S. Villar, L.A. Escaleira, Response sur-face methodology (RSM) as a tool for optimization in analytical chemistry, Talanta76 (2008) 965–977.

[44] J. Zolgharnein, A. Shahmoradi, J.B. Ghasemi, Comparative study of Box-Behnken,central composite, and Doehlert matrix for multivariate optimization of Pb (II)adsorption onto Robinia tree leaves, J. Chemom. 27 (2013) 12–20.

[45] A. Niazi, N. Khorshidi, P. Ghaemmaghami, Microwave-assisted of dispersive liquid-liquid microextraction and spectrophotometric determination of uranium afteroptimization based on Box-Behnken design and chemometrics methods,Spectrochim. Acta - Part A Mol. Biomol. Spectrosc. 135 (2015) 69–75.

[46] Y. Chen, Z. Sun, Y. Yang, Q. Ke, Heterogeneous photocatalytic oxidation of poly-vinyl alcohol in water, J. Photochem. Photobiol. A: Chem. 142 (2001) 85–89.

[47] W. Jiang, J.A. Joens, D.D. Dionysiou, K.E. O’Shea, Optimization of photocatalyticperformance of TiO2 coated glass microspheres using response surface metho-dology and the application for degradation of dimethyl phthalate, J. Photochem.

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6010

Photobiol. A: Chem. 262 (2013) 7–13.[48] Arnold E. Greenberg, R.Rhodes Trussell, Lenore S. Clesceri, Mary Ann H. Franson,

Water Pollution Control Federation; Standard methods for the examination of waterand wastewater, American Public Health Association, Washington, D.C, 2018 16thEdition. c1985, American Water Works Association.

[49] E. Haque, V. Lo, A.I. Minett, A.T. Harris, T.L. Church, Dichotomous adsorptionbehaviour of dyes on an amino-functionalised metal–organic framework, amino-MIL-101(Al), J. Mater. Chem. A 2 (2014) 193–203.

[50] B.H. Dang Son, V. Quang Mai, D. Xuan Du, N. Hai Phong, D. Quang Khieu, A studyon astrazon black AFDL dye adsorption onto Vietnamese diatomite, J. Chem. 2016(2016).

[51] P. Burroughs, A. Hamnett, A.F. Orchard, G. Thornton, Satellite structure in the X-ray photoelectron spectra of some binary and mixed oxides of lanthanum andcerium, J. Chem. Soc. Dalt. Trans. (1976) 1686.

[52] V. Fernandes, P. Schio, A.J.A. de Oliveira, W.A. Ortiz, P. Fichtner, L. Amaral,I.L. Graff, J. Varalda, N. Mattoso, W.H. Schreiner, D.H. Mosca, Ferromagnetisminduced by oxygen and cerium vacancies above the percolation limit in CeO2, J.Phys. Condens. Matter 22 (2010) 216004.

[53] C. Gionco, M.C. Paganini, S. Agnoli, A.E. Reeder, E. Giamello, Structural andspectroscopic characterization of CeO2–TiO2 mixed oxides, J. Mater. Chem. A 1(2013) 10918.

[54] B. Choudhury, P. Chetri, A. Choudhury, Oxygen defects and formation of Ce 3+

affecting the photocatalytic performance of CeO 2 nanoparticles, RSC Adv. 4 (2014)4663–4671.

[55] W. Zhu, S. Xiao, D. Zhang, P. Liu, H. Zhou, W. Dai, F. Liu, H. Li, Highly efficient andstable Au/CeO 2 –TiO 2 photocatalyst for nitric oxide abatement: potential appli-cation in flue gas treatment, Langmuir 31 (2015) 10822–10830.

[56] Y. Wang, J. Zhao, T. Wang, Y. Li, X. Li, J. Yin, C. Wang, CO2 photoreduction withH2O vapor on highly dispersed CeO2/TiO2 catalysts: surface species and their re-activity, J. Catal. 337 (2016) 293–302.

[57] F.W. Clarke, B. Irving Langmuir, Constitution of solids and liquids, J. Am. Chem.Soc. 38 (1916) 2221–2295.

[58] T.M.F. Marques, O.P. Ferreira, J.A.P. Da Costa, K. Fujisawa, M. Terrones,B.C. Viana, Study of the growth of CeO22 nanoparticles onto titanate nanotubes, J.Phys. Chem. Solids 87 (2015) 213–220.

[59] J. Jiao, Y. Wei, Z. Zhao, J. Liu, J. Li, A. Duan, G. Jiang, Photocatalysts of 3D orderedmacroporous TiO 2 -supported CeO 2 nanolayers: design, preparation, and theircatalytic performances for the reduction of CO 2 with H 2 O under simulated solarirradiation, Ind. Eng. Chem. Res. 53 (2014) 17345–17354.

[60] D. Tomova, V. Iliev, A. Eliyas, S. Rakovsky, Promoting the oxidative removal rate ofoxalic acid on gold-doped CeO2/TiO2 photocatalysts under UV and visible lightirradiation, Sep. Purif. Technol. 156 (2015) 715–723.

[61] J. Graciani, J.J. Plata, J.F. Sanz, P. Liu, J.A. Rodriguez, A theoretical insight into thecatalytic effect of a mixed-metal oxide at the nanometer level: the case of the highlyactive metal/CeOx/TiO2(110) catalysts, J. Chem. Phys. 132 (2010) 104703.

[62] V.V. Atuchin, V.G. Kesler, N.V. Pervukhina, Z. Zhang, Ti 2p and O 1s core levels andchemical bonding in titanium-bearing oxides, J. Electron Spectros. Relat.Phenomena 152 (2006) 18–24.

[63] Z. Zhang, Z. Zhou, S. Nie, H. Wang, H. Peng, G. Li, K. Chen, Flower-like hydro-genated TiO2(B) nanostructures as anode materials for high-performance lithiumion batteries, J. Power Sources 267 (2014) 388–393.

[64] X. Lu, X. Li, J. Qian, N. Miao, C. Yao, Z. Chen, Synthesis and Characterization ofCeO 2 / TiO 2 Nanotube Arrays and Enhanced Photocatalytic OxidativeDesulfurization Performance 661 (2016), pp. 363–371.

[65] C. Hao, J. Li, Z. Zhang, Y. Ji, H. Zhan, F. Xiao, D. Wang, B. Liu, F. Su, Enhancementof photocatalytic properties of TiO2 nanoparticles doped with CeO2 and supportedon SiO2 for phenol degradation, Appl. Surf. Sci. 331 (2015) 17–26.

[66] C. Karunakaran, P. Gomathisankar, Solvothermal synthesis of CeO 2 –TiO 2 nano-composite for visible light photocatalytic detoxification of cyanide, ACS Sustain.Chem. Eng. 1 (2013) 1555–1563.

[67] Z. Shi, P. Yang, F. Tao, R. Zhou, New insight into the structure of CeO 2 –TiO 2mixed oxides and their excellent catalytic performances for 1,2-dichloroethane

oxidation, Chem. Eng. J. 295 (2016) 99–108.[68] B. Yuan, Y. Long, L. Wu, K. Liang, H. Wen, S. Luo, H. Huo, H. Yang, J. Ma, TiO2@ h-

CeO2: a composite yolk-shell microsphere with enhanced photodegradation ac-tivity, Catal. Sci. Technol. 6 (2016) 6396–6405.

[69] C. Ampelli, R. Passalacqua, C. Genovese, S. Perathoner, G. Centi, T. Montini,V. Gombac, P. Fornasiero, Solar energy and biowaste conversion into H2 on CuO x/TiO2 nanocomposites, Chem. Eng. Trans. 35 (2013) 583–588.

[70] L. Matějová, K. Kočí, M. Reli, L. Čapek, A. Hospodková, P. Peikertová, Z. Matěj,L. Obalová, A. Wach, P. Kuśtrowski, A. Kotarba, Preparation, characterization andphotocatalytic properties of cerium doped TiO2: on the effect of Ce loading on thephotocatalytic reduction of carbon dioxide, Appl. Catal. B Environ. 152–153 (2014)172–183.

[71] X. Yang, L. Yang, J. Lin, R. Zhou, The new insight into the structure-activity relationof Pd/CeO 2 -ZrO 2 -Nd 2 O 3 catalysts by Raman, in situ DRIFTS and XRD Rietveldanalysis, Phys. Chem. Chem. Phys. 18 (2015) 3103–3111.

[72] A.I. Kontos, A.G. Kontos, D.S. Tsoukleris, G.D. Vlachos, P. Falaras,Superhydrophilicity and photocatalytic property of nanocrystalline titania sol–gelfilms, Thin Solid Films 515 (2007) 7370–7375.

[73] M. Guo, J. Lu, Y. Wu, Y. Wang, M. Luo, UV and visible Raman studies of oxygenvacancies in rare-earth-doped ceria, Langmuir 27 (2011) 3872–3877.

[74] A. Filtschew, K. Hofmann, C. Hess, Ceria and its defect structure: new insights froma combined spectroscopic approach, J. Phys. Chem. C 120 (2016) 6694–6703.

[75] T.M.F. Marques, O.P. Ferreira, J.A.P. da Costa, K. Fujisawa, M. Terrones, B.C. Viana,Study of the growth of CeO2 nanoparticles onto titanate nanotubes, J. Phys. Chem.Solids 87 (2015) 213–220.

[76] M.E. Contreras-Garc??a, M.L. Garc??a-Benjume, V.I. Mac??as-Andr??s, E. Barajas-Ledesma, A. Medina-Flores, M.I. Espitia-Cabrera, Synergic effect of the TiO2-CeO2nanoconjugate system on the band-gap for visible light photocatalysis, Mater. Sci.Eng. B Solid-State Mater. Adv. Technol. 183 (2014) 78–85.

[77] S. Luo, T.D. Nguyen-Phan, A.C. Johnston-Peck, L. Barrio, S. Sallis, D.A. Arena,S. Kundu, W. Xu, L.F.J. Piper, E.A. Stach, D.E. Polyansky, E. Fujita, J.A. Rodriguez,S.D. Senanayake, Hierarchical heterogeneity at the CeOx-TiO2 interface: electronicand geometric structural influence on the photocatalytic activity of oxide on oxidenanostructures, J. Phys. Chem. C 119 (2015) 2669–2679.

[78] Y. Xu, M.A.A. Schoonen, The absolute energy positions of conduction and valencebands of selected semiconducting minerals, Am. Mineral. 85 (2000) 543–556.

[79] J. Tian, Y. Sang, Z. Zhao, W. Zhou, D. Wang, X. Kang, H. Liu, J. Wang, S. Chen,H. Cai, H. Huang, Enhanced photocatalytic performances of CeO2/TiO2 nanobeltheterostructures, Small 9 (2013) 3864–3872.

[80] B.H.D. Son, V.Q. Mai, D.X. Du, N.H. Phong, N.D. Cuong, D.Q. Khieu, Catalytic wetperoxide oxidation of phenol solution over Fe–Mn binary oxides diatomite com-posite, J. Porous Mater. 24 (2017) 601–611.

[81] J. Saien, S. Khezrianjoo, Degradation of the fungicide carbendazim in aqueous so-lutions with UV/TiO2 process: optimization, kinetics and toxicity studies, J.Hazard. Mater. 157 (2008) 269–276.

[82] J.R. Kim, B. Santiano, H. Kim, E. Kan, Heterogeneous oxidation of methylene bluewith surface-modified iron-amended activated carbon, Am. J. Anal. Chem. 2013(2013) 115–122.

[83] M.A. Henderson, C.L. Perkins, M.H. Engelhard, S. Thevuthasan, C.H.F. Peden,Redox properties of water on the oxidized and reduced surfaces of CeO2(1 1 1),Surf. Sci. 526 (2003) 1–18.

[84] D.R. Mullins, P.M. Albrecht, T.-L. Chen, F.C. Calaza, M.D. Biegalski, H.M. Christen,S.H. Overbury, Water dissociation on CeO 2 (100) and CeO 2 (111) thin films, J.Phys. Chem. C 116 (2012) 19419–19428.

[85] T.K. Jana, A. Pal, K. Chatterjee, Self assembled flower like CdS-ZnO nanocompositeand its photo catalytic activity, J. Alloys. Compd. 583 (2014) 510–515.

[86] a Houas, Photocatalytic degradation pathway of methylene blue in water, Appl.Catal. B Environ. 31 (2001) 145–157.

[87] K. Mehrotra, G.S. Yablonsky, A.K. Ray, Macro kinetic studies for photocatalyticdegradation of benzoic acid in immobilized systems, Chemosphere 60 (2005)1427–1436.

L.T.T. Tuyen et al. Journal of Environmental Chemical Engineering 6 (2018) 5999–6011

6011